Detailed Analysis
Agensi, launched by Netherlands-based non-technical founder Samuel and built primarily using Claude Code and Lovable, represents an early attempt to bring curation, security, and monetization infrastructure to the rapidly emerging ecosystem of AI agent skills. The platform centers on SKILL.md, a folder-plus-instructions format introduced by Anthropic to enable AI coding agents — including Claude Code, Cursor, Codex CLI, and Gemini CLI — to be taught new, reusable capabilities. With 200+ skills across categories ranging from DevOps and deployment to productivity, and approximately 7,000 active users over the preceding 28 days, Agensi occupies an early but not yet dominant position in a marketplace segment that has grown quickly, with competing platforms such as LobeHub Skills (169,739 skills) and agentskill.sh (110,000+) operating at significantly larger scale.
The platform's most substantive differentiator is its emphasis on security vetting, a concern grounded in documented research. Earlier in 2026, studies including the ToxicSkills and ClawHavoc reports found that roughly 36% of sampled SKILL.md files contained prompt injection vectors — a meaningful threat given that skills execute with direct access to a user's local machine environment. Agensi's automated scanning layer checks for permission boundary violations, suspicious outbound network requests, problematic dependencies, and common malware signatures before any skill goes live. The founder acknowledges its limitations, positioning it as a filter for obvious threats rather than a replacement for individual code review, and leaves open questions around creator identity verification and public scan report transparency — decisions that will significantly shape the platform's trust profile as it scales.
The monetization architecture Agensi has constructed attempts to solve a real gap in the agent-tooling ecosystem: compensating developers who invest time encoding domain expertise into reusable workflows. Creators choose between direct sales, retaining 80% of revenue after a 20% platform cut plus a $0.50 per-sale fee, and a subscription pool model where 70% of MCP subscription revenue is distributed monthly based on actual usage pull. Both streams can run simultaneously. With only $200 in gross merchandise volume in its first month and a single paying MCP subscriber, the commercial traction is nascent, but the structure reflects deliberate thinking about aligning creator incentives with platform health — a challenge that app stores and plugin marketplaces have historically struggled to solve equitably.
The platform's strategic vulnerability lies in its platform dependency and competitive ceiling. Anthropic, Cursor, and other agent tooling companies could launch first-party skill stores at any point, leveraging existing user bases and native integration advantages. Samuel's stated thesis — that a curated, cross-agent marketplace is defensible precisely because vendor-owned stores will always be agent-locked — is plausible but unproven. Cross-platform compatibility could attract creators who want broad distribution, but it may also mean Agensi lacks the deep integration hooks that first-party stores can offer. The MCP server component, which gives agents four native tools to search and retrieve skills mid-task without requiring a user to manually browse, is Agensi's clearest attempt to embed itself in agentic workflows in a way that is harder for incumbents to replicate without deliberate effort.
Agensi's emergence reflects a broader structural shift in how AI capability is being packaged and distributed. As large language model agents move from chat interfaces to autonomous coding and task-execution environments, the software supply chain question — who builds agent capabilities, who vets them, and how developers are compensated — becomes increasingly consequential. The SKILL.md format itself, and the ecosystem of marketplaces forming around it, mirrors earlier dynamics seen in plugin stores for IDEs and browsers: initial fragmentation, followed by consolidation around platforms with the strongest curation, distribution, and trust signals. Whether Agensi can establish the trust and creator relationships necessary to anchor that consolidation, before better-resourced entrants move decisively, is the central open question its early metrics have yet to answer.
Read original article →